The system you actually have
Most growing ISPs run on a stack nobody chose: a spreadsheet of subscribers, M-Pesa messages forwarded to a WhatsApp group, a MikroTik only one person can log into, and an end-of-month ritual that eats three days. It works — right up to the subscriber count where it doesn't. Migration isn't about better software so much as getting the business out of individual people's heads.
What moves, in what order
The sequence that avoids downtime is always the same: data first, money second, enforcement last.
- Data: subscribers, plans and balances come in through the streamed bulk import — thousands of rows, validated per row, built for cutover day
- Money: Daraja is pointed at your paybill; payments start reconciling automatically while the old process still runs in parallel
- Enforcement: only when imports are verified and payments are flowing does suspend/restore go live — the moment billing state starts controlling the network
Data first, money second, enforcement last.
The parallel-running week
For the first week, run both: the old spreadsheet stays open while CoreSync does the work. Every evening, compare — did the platform see every payment the phone saw? Discrepancies in this week are cheap; they're how the team learns to trust the meter before the meter takes over. When two consecutive days reconcile perfectly, the spreadsheet retires.
What changes for subscribers: nothing, then everything
Done right, subscribers notice nothing on migration day — same plans, same paybill, same connectivity. What they notice over the next month is the platform: reminders before expiry instead of surprise disconnections, reconnection in seconds after paying, and a portal where they see their own account. Churn responds to that experience more than to any speed upgrade.